engelresearch

PORTFOLIO DESK

Research Principles and System Methodology

The foundational document for Engel Research’s research environment, portfolio construction framework, and decision-making process.

engel-research.com

Why Portfolio Desk Was Built

We believe the capital markets should be approached as a business. As with any successful business, long-term success does not result from random decisions, but from a consistent, measurable process that can be repeated over time.

For a process to remain consistent, it must be grounded in reliable research that reflects real-world conditions. Its conclusions should be applicable repeatedly, without constant modification and without reliance on emotional decision-making.

Over the years, algorithmic trading systems have become increasingly complex. Dozens of indicators, hundreds of parameters, and countless calibration options can produce impressive backtests, yet often fail to reproduce the same behavior in live trading.

Many existing systems also operate as a black box. The user receives an output without understanding how it was generated, which assumptions underpin it, or how dependable it may be over time.

Portfolio Desk was created from the belief that the key to success is not the most complex algorithm, but the most reliable research process.

The system was built around a simple principle: one integrated environment that brings together the entire workflow — research, portfolio construction, and independent portfolio management.

During development, we worked to reduce research bias, eliminate unnecessary complexity, and build a process suited to changing market conditions. Simplicity is not a compromise; it is a prerequisite for a methodology designed to endure.

What Portfolio Desk Is Not

Portfolio Desk is a versatile platform that combines research tools, portfolio management capabilities, and ongoing monitoring and oversight. Its limitations are as important to understand as its capabilities.

It Does Not Guarantee Returns

The system is designed to support a consistent research process, a repeatable operating framework, the construction of investment and trading strategies, and ongoing portfolio management. No system can guarantee returns, and all capital-market investments involve risk.

It Does Not Predict the Market

The system does not attempt to forecast events or predict market direction. It enables systematic rotation among selected assets based on relative strength and momentum. Combined with diversified portfolio construction in a single environment, this approach treats investing as an operating process rather than a collection of isolated trades.

It Is Not Built for Intraday Trading

Decisions are made after the market closes and implemented on the following trading day. This is intended to reduce the gap between research assumptions and real execution conditions, while filtering intraday noise. The system evaluates stocks and ETFs over time horizons that allow trends to develop rather than reacting to every short-term fluctuation.

It Is Not a Black Box

The methodology, research process, ranking model, and test results are presented transparently. All results displayed in the system are based on robustness testing, not optimization results alone.

It Does Not Replace Judgment

Portfolio Desk provides a research environment, portfolio-management tools, and decision-support capabilities. Responsibility for capital allocation, risk tolerance, and system operation remains with the trader or investor.

The Challenges of Algorithmic Research

The purpose of algorithmic research is not to produce the most impressive equity curve, but to estimate how a strategy is likely to behave in live trading. In practice, a meaningful gap often exists between backtest results and real-time performance.

One of the primary causes is overfitting. During research, parameters are repeatedly calibrated to achieve the best historical result. As the number of parameters and calibration cycles increases, so does the probability that the model is fitting past data rather than identifying principles that are more likely to persist.

A second challenge is the proliferation of indicators and parameters. Every additional variable substantially expands the number of possible combinations, increasing the likelihood that the algorithm selects a configuration that appears optimal only in hindsight. The larger the search space, the lower the probability that the selected combination will reproduce itself under real market conditions.

Research design itself may also introduce bias. Many systems treat intraday or closing prices as though a trade could have been executed at that exact moment. In practice, the decision is often available only after the close, while execution occurs on the next trading day at a different price. This gap can materially distort the relationship between research results and live performance.

Research should approximate reality as closely as possible, rather than optimize for statistical perfection on historical data.

Design Principles

Portfolio Desk was not built to identify the most profitable strategy on historical data. It was built to provide a reliable, simple, and systematic research environment capable of generating investment decisions that can be implemented under real trading conditions.

Simplicity Before Complexity

As the number of research parameters increases, so does the likelihood of accidental fit to historical data. The system therefore relies on a limited set of carefully selected parameters, each of which must demonstrate a meaningful contribution. The objective is not to build the most complex model, but the model most likely to remain dependable over time.

Robustness Before Return

Many research processes select the parameter set that produced the highest return. Portfolio Desk follows a different approach. Optimization is only the first stage. Once the leading parameter set is selected, it undergoes a robustness test to determine whether the same parameters continue to behave consistently across different market periods.

Research That Reflects Real-World Execution

The system does not rely on assumptions that cannot be implemented in live trading. Decisions generated during or after the trading session are not executed immediately, but on the following trading day. This principle reduces the gap between research results and actual execution conditions, while allowing an asset to develop with its trend without overreacting to intraday noise.

A Portfolio, Not an Isolated Trade

Rather than attempting to predict the next move in a single stock, Portfolio Desk is built around the research and management of a diversified portfolio. The system ranks more than one hundred financial assets each day, enabling users to construct portfolios that adapt to changing market conditions.

Full Transparency

Research, rankings, executions, and system results are displayed transparently. The objective is not to create a black box, but to allow users to understand how decisions were reached and the principles on which they are based.

Robustness — The system performs optimization over a defined research period. After the leading parameter set is selected, the parameters are frozen and reapplied to each research year without re-optimizing each year independently. The purpose is to evaluate the stability of the same parameter set across different market regimes and reduce reliance on a single optimization result. This is not a full walk-forward test, but an annual robustness framework designed to assess whether the parameters remain consistent over time.

The Research Environment

5.1 Asset Universe

The capital markets include tens of thousands of stocks and ETFs. Rather than attempting to monitor an unlimited number of instruments, Portfolio Desk operates through a defined asset universe built through extensive research to represent a broad range of market conditions.

The asset universe includes approximately 110 carefully selected financial instruments. Around 70 are ETFs covering the 11 S&P 500 sectors, industry groups, bonds, gold, silver, copper, Bitcoin, agriculture, geographic markets, and other complementary assets. It also includes approximately 40 leading S&P 100 stocks representing all major sectors.

Each asset is studied across an extended historical period that includes multiple market regimes: inflationary environments, geopolitical crises, growth periods, sharp declines, sector rotation, and other changing conditions.

A defined asset universe provides full control over the research environment. Heat maps and the ranking system make it possible to identify capital flows across sectors and assets, monitor market rotation, or allow the system to operate automatically according to predetermined research rules.

בנק הנכסים
The asset universe displays all financial instruments included in the system, their rankings, and their classification by sector, complementary asset class, and geographic market.

5.2 Ranking System

The ranking system was developed to reduce the number of indicators required for decision-making and consolidate several essential components into a single primary score.

Each asset receives a score from 0 to 10, calculated from four primary components: relative strength versus the market, asset momentum, distance from the 200-day moving average, and the asset’s relative position within the broader universe.

These four components are combined into a consistent daily ranking that supports the research process, portfolio construction, and selection of active positions.

מערכת הדירוג
The ranking system combines four core components — relative strength, momentum, distance from the 200-day moving average, and relative position within the asset universe — into a single score from 0 to 10.

5.3 Optimization Engine

Every symbol considered for inclusion in a portfolio first undergoes a research process through the optimization engine.

The process evaluates a limited set of predefined parameters to identify an appropriate operating configuration for each symbol. Unlike algorithmic systems that test hundreds of thousands or millions of possible combinations, the optimization engine works within a constrained and clearly defined research space, typically consisting of hundreds to several thousand combinations.

This approach enables fast research, often completed within seconds. The purpose is not merely to shorten computation time, but to reduce the risk of overfitting — where the algorithm conforms to historical data rather than identifying principles with a higher probability of recurring.

At the end of the optimization process, each symbol receives a dedicated parameter set. This configuration defines how the asset operates within the system and becomes the basis for the next stage: stability testing through the robustness framework.

מנוע האופטימיזציה
The optimization engine defines the research ranges, evaluates a limited parameter set, and produces the configuration used in subsequent robustness testing.

5.4 Robustness Testing

Optimization is the starting point of the research process, not its conclusion.

After the leading parameter set is selected, the parameters are frozen. The same configuration is then retested across each research year without any further optimization.

The objective is to determine whether the same parameters continue to behave consistently across different market regimes. This materially reduces the risk of research bias and dependence on a single favorable result.

The rationale is straightforward. Every optimization process introduces some degree of bias. When an algorithm searches for the parameter set with the best historical result, the selected outcome may reflect a coincidental fit rather than a durable investment principle. Robustness testing reduces this risk by evaluating the same parameters across different market periods without further adjustment.

מחקרי Robustness
The robustness screen displays the performance of the frozen parameter set, with each research year evaluated independently and without further optimization. The purpose is to assess parameter stability across different market cycles.

5.5 Portfolio Construction

Once the research stage is complete, assets move into the portfolio-construction environment, where a wide range of investment strategies can be built according to the investor’s objectives. These may include equity portfolios, ETF portfolios, growth strategies, energy portfolios, geographic-market portfolios, alternative-asset portfolios, and combinations among them.

The system includes a correlation tool that measures relationships among assets and helps identify combinations with relatively low correlation, with the aim of reducing volatility and drawdown. It also presents annual-return data, portfolio composition, and information that helps identify assets that may detract from overall performance.

From the assets selected for a portfolio, the system activates only the highest-ranked instruments at any given time. For example, if a portfolio contains five assets but only three may be active simultaneously, the system continuously selects the three highest-ranked assets, subject to the entry and exit rules defined for each symbol during research.

The portfolio is not static. It is updated continuously according to asset rankings and predefined research rules.

בניית הפורטפוליו
The portfolio-construction environment supports multiple portfolio types, correlation analysis, limits on the number of active assets, and ongoing management of portfolio composition according to predefined rules.

5.6 Signal System

After portfolio construction is complete, up to three portfolios can be transferred into active trading mode.

Each portfolio displays a real-time performance chart, monthly returns, a trade log, and buy and sell signals. Through third-party software, the system may also be connected to an automated trading platform so that orders are executed according to system signals without manual intervention.

מערכת האיתותים
The signal system displays model instructions as a portfolio moves from research into active trading, including the symbol, order type, position status before and after execution, and entry date.

Portfolio Construction

The Portfolio Desk philosophy is not centered on finding the next trade, but on building a complete investment portfolio.

Each asset undergoes independent research, optimization, and robustness testing before becoming eligible for inclusion in a portfolio. Once research is complete, stocks and ETFs can be combined, their correlations evaluated, and investment strategies constructed according to the investor’s objectives and desired risk profile.

The portfolio approach evaluates each asset as part of a broader system. Buy and sell decisions consider portfolio composition, asset rankings, and relationships among holdings rather than isolating a single trade or reacting to every short-term market movement.

Our Vision

We believe the capital markets are neither a game nor a wager. They should be approached as a business, and success — as in any successful business — does not come from luck, intuition, or one exceptional trade, but from a consistent, measurable, repeatable process.

Our vision is to give independent investors access to the same research discipline that was once available mainly to professional institutions. Not through promises of return, but through a clear methodology, a structured operating process, and the ability to make consistent decisions over time.

We believe the quality of decisions matters more than the number of decisions. We therefore do not encourage chasing the next trade or short-term success. A diversified portfolio grounded in research, risk management, and discipline is the appropriate way to navigate changing markets over time.

Portfolio Desk was built with one objective: to replace emotional decision-making with a systematic operating process — making research the foundation, discipline the habit, and the portfolio a unified system governed by clear rules.

If more investors begin to treat the capital markets as a business, build a process instead of chasing forecasts, and act consistently rather than emotionally, we will have achieved our purpose.

We do not believe the market can be predicted. We believe there is a disciplined way to operate within it.